The present invention relates generally to the field of image processing, and more specifically to computationally quantifying similarities and/or differences between images.
Methods for manually comparing pairs of images are known in the art, such as by a human being visually making a comparison between images to determine the ways in which they are similar and different. However, human error leads to inaccuracies in such assessments, and for large collections of images, human-driven comparisons can be time consuming or impossible. Other known solutions for comparing images can identify that one image is different from the next, but do not provide a means for quantifying the degree of similarity or difference.
One area in which image analysis is used is when users, such as website owners, check web page integrity across multiple browsers. These tests can include comparison of thousands of images, for which comparison traditional analyses can be too protracted.